TIS-100 Bench: Image Test Pattern 2

Draw a second fixed target image. The task is to encode its spatial pattern within the small programs available on the grid.

In TIS-100 Bench, 16 LLMs each had one attempt at this TIS-100 puzzle, with 10 minutes to find a working program and keep improving it, without seeing the community leaderboard. 14 of 16 models solved it. No model found a new Pareto-frontier point beyond the community leaderboard.

14 / 16models solved it
1,151best LLM cycles (1,150 community)
0new frontier points

Records

ScoreCommunity bestBest LLMModel
Cycles1,1501,151Gemini 4 Argon · +0%
Nodes11Fable 5.1 Ties record
Instructions911Fable 5.1 · +22%

Community records from the community leaderboard when we froze it. A record counts only after fresh-game replay and 100,000 passing random tests.

Pareto frontier

Pareto explorer

Lower is better ↙
Loading interactive Pareto explorer…
Community leaderboardModel frontier pointNew frontier point
Select a point with the mouse or keyboard to inspect its scores and program.

Search over time

Loading puzzle trajectories…

Every model

ModelSolvedBest cyclesBest nodesBest instructionsNew frontier points
01Gemini 4 ArgonYes1,151211—
02Opus 5.5Yes1,151212—
03Sonnet 5.5Yes1,151218—
04GPT-6 AstraYes1,153113—
05GPT-6.1 SolYes1,169111—
06Fable 5.1Yes1,206111—
07Muse Wonder 1.0Yes1,302212—
08Gemini 3.8 FlashYes1,381212—
09Haiku 5.5Yes1,382218—
10Muse Spark 1.3Yes1,399212—
11Kimi K3.1 ProYes1,400215—
12GPT-6 LunaYes1,415316—
13Grok 4.7Yes1,419219—
14DeepSeek V4.1 FlashYes1,470326—
15GLM-5.3No————
16MiMo-V2.6-ProNo————

Best scores can come from different programs. New frontier points are not matched or beaten on all three scores by any community program.

Best model programs

These programs are in TIS-100 save format and released under the MIT License. All of this puzzle’s model programs are on GitHub and Hugging Face.

Sonnet 5.5 Fewest cycles

1,151 cycles · 3 nodes · 21 instructions51781.1151-3-21.txt on GitHub

Sonnet 5.5 program running on 51781 in TIS-100
Node 5
MOV 0, DOWN
MOV 15, ACC
P: MOV 3, DOWN
SUB 1
JGZ P
MOV 1, DOWN
MOV 15, ACC
Q: MOV 3, DOWN
SUB 1
JGZ Q
Node 8
MOV ACC, RIGHT
SWP
MOV 14, ACC
P: MOV 0, RIGHT
SUB 1
JGZ P
MOV -1, RIGHT
SWP
ADD 1
Node 9
MOV UP, DOWN
MOV LEFT, DOWN

GPT-6.1 Sol Fewest nodes

1,490 cycles · 1 nodes · 15 instructions51781.1490-1-15.txt on GitHub

GPT-6.1 Sol program running on 51781 in TIS-100
Node 9
NEG
MOV ACC DOWN
SWP
MOV ACC DOWN
ADD 1
SWP
ADD 27
MOV 3 DOWN
L: MOV 0 DOWN
MOV 3 DOWN
MOV 0 DOWN
MOV 3 DOWN
SUB 4
JGZ L
MOV -1 DOWN

GPT-6.1 Sol Fewest instructions

1,758 cycles · 2 nodes · 11 instructions51781.1758-2-11.txt on GitHub

GPT-6.1 Sol program running on 51781 in TIS-100
Node 8
MOV ACC RIGHT
ADD 1
Node 9
NEG
MOV ACC DOWN
MOV LEFT DOWN
ADD 29
L: MOV 3 DOWN
MOV 0 DOWN
SUB 2
JGZ L
MOV -1 DOWN